You are on page 1of 16

Cement and Concrete Research 64 (2014) 7388

Contents lists available at ScienceDirect

Cement and Concrete Research


journal homepage: http://ees.elsevier.com/CEMCON/default.asp

Effects of the presence of free lime nodules into concrete:


Experimentation and modelling
Luc Courard , Herv Dege, Anne Darimont
University of Lige, Department of Architecture, Geology, Environment and Constructions, Belgium

a r t i c l e

i n f o

Article history:
Received 1 August 2013
Accepted 9 June 2014
Available online 12 July 2014
Keywords:
Hydration (A)
Mechanical properties (C)
CaO (D)
Concrete (E)
Pop-out

a b s t r a c t
When nodules of lime are embedded into concrete, the expansion accompanying the transformation of CaO into
Ca(OH)2 induces stresses and strains in both the lime nodule and in the concrete matrix. The concrete cover
thickness, the diameter and the shape of the lime nodule as well as the mechanical characteristics of concrete
and lime are the key parameters inuencing the development of internal pressure and hence controlling the
risk of cracking or pop-out. In order to study the effect of lime into cementitious concretes, laboratory investigations and modelling have been performed and show that the minimum cover thickness necessary to avoid the
development of the pop-out phenomenon is estimated of the order of half the diameter of the inclusion. This
is coming from the observation that expansion happens inside the porosity of the hydrated lime Ca(OH)2:
ESEM and DRX analyses conrm the effect of connement in the development of crystals.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
Lime has been used for a very long time in construction and buildings: Roman cement was already made of a part of lime while it
remained the only binder used until modern cement was designed
during XIXth century [1]. Lime is an industrial product obtained by
calcination of limestone in a lime kiln [2]. This is described as a bright
lime (Table 1), because of its high reactivity with water. The bulk density
of the limestones industrially used for the manufacture of lime usually
offers a lower density than calcite used for ornamental stones: porosity
maybe up to 30% [3]. Quick lime is very reactive with water and
hydrates are quickly formed [4]. Hydration process is accompanied by
a signicant proliferation (Table 1). The formation of Ca(OH)2 yields
in larger volume (expansive reaction). The ratio of volume change
from CaO particle to Ca(OH)2 is 33.1/16.8 2.
The doubling of the molar volume (from 16.8 to 33.1 cm3/mol) is responsible for expansion during hydration [4]. The intensity and speed of
hydration are governed by lime purity, particle size, surface area, etc.
[5]. Burning temperature and kiln technology are also two important
discriminant factors in the case of industrial lime production [3].
An interesting parameter used to quantify the reactivity of lime is
the so-called T60, which is measured in accordance with standardized
method EN 459-2:2001. It gives the speed of lime extinction, or the
Corresponding author at: University of Lige, Civil and Mechanical Engineering
Institute (B 52/3). Chemin des Chevreuils, 1, B-4000 LIEGE, Belgium. Tel.:+32 4 366 93
50; fax: +32 4 366 92 38.
E-mail address: Luc.Courard@ulg.ac.be (L. Courard).

http://dx.doi.org/10.1016/j.cemconres.2014.06.005
0008-8846/ 2014 Elsevier Ltd. All rights reserved.

time needed to attempt a temperature of 60 C: the smaller it is, the


more reactive the lime is. In some cases, the lime can be dead burned,
leading to high density CaO grains [4]. This dead burn lime hydrates
very slowly because of a reduced porosity [6]. When lime is incorporated into concrete, problems due to expansion may occur (Fig. 1): this
phenomenon is well known as pop-out [711].
In many cases, quick lime is present in steel or iron slags [8,9]. That
means that it is rarely in the form of millimetre-sized aggregate in a conned environment. In the case of steel slags, Deng et al. [10] observed
that expansion rates are depending on the type of cement and the percentage of lime: for lime contents of 2 and 5% (by weight of cement), the
maximum observed rate of expansion is 0.12 and 0.7%, respectively, for
cement type CEM I. The expansion force is estimated at 11.87 MPa in
3 days. A dead lime was used for experimentation and required alkali
activation: the concentration of OH-ions in the pore solution of cement
paste controls the expansion by affecting the positions occupied by the
crystals of Ca(OH)2 and the pressure of crystallization. Analyses of the
behaviour of LD steel slags containing lime nodules [11] were also conducted as a result of damage observed. An expansion rate of 0.16%
(measured by immersion according to the Korean standard KS F 2580)
was considered. The nite element calculations show that the depth of
the pop-out increases as concrete strength decreases and the diameter
of the slag increases (Fig. 2).
Other authors [12,13] made investigations on ne lime particle coming from shrinkage preventing agents or wrong cement manufacturing.
Useful information may be found through the study of other
expanding processes [14]: Alkali Aggregate Reaction (AAR) should induce similar stresses inside concrete. A difference is however coming

74

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

Table 1
Hydration of quick to hydrated lime [2,3].
Property

CaO

+H2O

Ca(OH)2

Molecular weight
Bulk density (g/cm3)
Specic density (g/cm3)
Molar volume (cm3/mol)

56.08
1.401.90
3.33
56.08/3.33 = 16.8

18.01

74.09
0.450.65
2.24
74.09/2.24 = 33.1

from the easier fullling of aggregate cracks by the silica gel which
progressively replaces a part of the initial products present along the
edge of stone material.
With regard to the very poor information coming from literature
reviewing, it clearly appears a lack of knowledge on the behaviour of
quick lime aggregates (up to 20 mm diameter) when mixed into
concrete. A risk evaluation analysis is needed: it will be based on an experimental programme and modelling that will help in understanding
free lime behaviour in conned situation.

Fig. 2. Concrete cover versus diameter of the slag for different concrete types (from [11]).

It is experimentally observed that K is about 4.9 for ordinary


concrete [16]. Bolomey formula also suggests a linear relationship
between the compressive strength and the ratio C/W.

2. Stresses calculation and modelling: theoretical background


f c;cube k

2.1. Simplied approach


As a rst simplied preliminary approach, it is considered that the
swelling pressure is unable to induce cracking in the concrete as a
nodule of lime can merely be considered as an air bubble, whose effect
is increasing the porosity and, consequently, decreasing the compressive strength of concrete. Bolomey and Feret theories can be used to
quantify this phenomenon [15].
Based on the equation p (aggregates) + s (sand) + c (cement) + w
(water) + v (voids) = 1, which expresses the sum of the volume fractions for 1 m3 of concrete, we have, with = (c/(c + w + v)), the Feret
formula which expresses the relationship between the compressive
strength of concrete and the voids (v):
2

f c;cube K0 K0

2
c
with K0 K  Rc
cwv

where fc,cube is the compressive strength (MPa), K is a granular


coefcient and Rc = the compressive strength of cement measured on
standardized mortar (EN 196-1).
This clearly indicates that the compressive strength decreases when
the w/c ratio increases. If we express this equation as a function of W
and C (mass ratio) for a cement relative density of 3.15, the expression
can be written:
1
f c;cube K0  
2 :
1 3:15 CW
V

1cm

C
h1
W

where k (2636) and h1 (0.45 to 0.87) depend on the quality of the


cement, the age of the concrete, the shape and dimensions of the test
pieces, the curing conditions and the sieving curve of aggregates and
sand.
The volume occupied by the nodules can be considered as an additional volume of water in the sense of increasing W/C ratio; that will
partially produce an additional volume of air after curing and evaporation. If we consider for example a W/C = 0.5 and a bulk density of
lime 1.56 [3], the volume occupied by the nodules, for a percentage of
0.3% of the mass of aggregates into concrete (1300 kg/m3 of concrete),
would be: 0.003 1300/1560 = 0.0025 m3 = 2.5 l. This means that
we can consider a ctitious increase of the amount of water for a
350 kg concrete cement of about 2.5 l, which means a total of
175 + 2.5 = 177.5 l. The W/C ratio increases thus from 0.50 to 0.507.
Feret formula allows estimating the resulting loss of strength [16]:
1
1 3:15  0:502

1
1:017:
1 3:15  0:512

This corresponds to a loss of strength of about 1.7%.


This evaluation clearly shows that the inuence of nodules inside the
concrete has a very marginal impact on the major structural characteristic of concrete: compressive strength is only lightly affected by a
reasonable level of pollution by lime nodules.
However, if these nodules are close to the surface, they are likely
to induce pop-out and cracking, which is detrimental for concrete

2cm

Fig. 1. Degradations induced by pop-out in concrete.

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

75

blocks [11]. These slags contain mostly dead-burned lime nodules,


much slower to react because these are more compact than the type
of lime considered here.
Specic theoretical developments have thus been carried out to
assess the risk of such phenomena. Cracking and the emergence of a
burst are indeed conditioned by a number of factors related to the
materials:

the depth of the nodule,


the diameter of the nodule,
the concentration of nodules,
the conditions of connement of the nodule of lime (rate of
expansion, swelling pressure),
tensile strength of the concrete,
modulus of rigidity of lime (and hydrated lime) and concrete.

Fig. 3. General principle of the model of embedded nodule.

For this purpose, simple theoretical models are proposed. These


models are based on the mechanics of materials and on the theory of
elasticity. The study also assesses the sensitivity of the mechanical effects with respect to the various relevant physical parameters. Two congurations are studied. The rst one considers a nodule embedded
within the concrete mass: the objectives are in this case to estimate
the inuence zone of the nodule and the possible risks of interaction between neighbouring nodules, as well as the risk of crushing or cracking
of the concrete in the vicinity of the nodule. The second situation is
considering a nodule located near the free surface of the concrete,
with the main objective of evaluating the risk of occurrence of a popout phenomenon.

Fig. 4. Principle of concrete rupture due to near-to-surface swelling nodule.

2.2. Modelling of expansive nodule inside a rigid medium


structure durability [17]. Only few data are available on this subject in
the literature [10,12,18] and mainly deals with the effect of the nodules
of lime in steel slag, most often used as aggregates for making concrete

2.2.1. Behaviour of a nodule embedded within concrete


Model 1 used in this situation considers a spherical nodule in perfect
contact with an environment encompassing innite dimensions. Both

Fig. 5. Preparation of the concrete blocks.

76

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

media are assumed to exhibit an elastic behaviour. Although both materials are known to be largely inelastic, this model is however appropriate for low level of stresses and in particular to estimate the initiation
of cracking and can also provide a good understanding of the physical
phenomena with a limited computational effort.
Model 1 (Fig. 3) is based on the assumption that the swelling of the
nodule is partially prevented by pressure developing at the interface
between concrete and lime and conning the nodule. This pressure depends on the mechanical properties (elastic modulus and Poisson's
ratio) of the two materials, as well as on the amplitude of the swelling
as it would be observed if the nodule was perfectly free to expand
upon hydration. Assuming a spherical symmetry of the problem and a
perfect compatibility of the displacements at the interface between
lime nodule and concrete matrix, the pressure can be calculated using
Eq. (3), on the base of the relations proposed in [19] for spherical
containers under internal or external uniform pressure:
R
R
p
CL CC

with
CL

1 2L
EL

D=2
r p

3
D
2z

and
12C
EC

s
3 Vfinal
1:
Vinitial

3
p D=2
D

3 :
2 z

where EL and EC are the elastic modulus of hydrated lime and concrete,
respectively; L and C are the Poisson's ratio of hydrated lime and
concrete, respectively; R is the initial radius of the nodule and R is
the variation of the radius of the nodule due to hydration process.
R/R ratio can be directly related to volume variation by means of
Eq. (6):
R

surrounding medium from the elasticity theory, as proposed in [19]. It


is needed to distinguish compressive stresses r, acting in radial direction, and tensile stresses t, oriented in the circumferential direction.
They are given in Eqs. (7) and (8), respectively:
3

and
CC

Fig. 7. Immersion of the concrete block into water.

If a perfect contact is assumed with a contact pressure that is fully


developed, it is possible to calculate the stress distribution in the

In these equations, the stresses are positive in tension and negative


in compression. Furthermore, D is the diameter of the inclusion and z
is the distance from the interface concrete/inclusion to the considered
point inside the concrete.
These stresses can then be compared to the compressive and tensile
strength of the concrete in order to assess the risk of a local crushing
around the nodule or of occurrence of radial cracks associated with
circumferential tension in the vicinity of the nodule. On the other
hand, the model allows estimating the area of concrete mechanically
disturbed by the presence of a swelling nodule.
2.2.2. Effect of a nodule in the vicinity of the concrete surface
A second model (Fig. 4) is used to assess the risk of local bursting of
concrete (pop-out). It is now assumed that the swelling pressure is no
more balanced by a stress distribution with spherical symmetry, but
rather by a tensile stress distribution varying linearly from the inclusion
towards the surface and distributed on a truncated cone.
Based on this failure pattern, one can evaluate the maximum tensile
stress acting at the base of the truncated cone with the following
equation:
2

pR
#
t "
2
R e
RR e

tg
3tg2

where R is the radius of the nodule, p is the pressure induced at the interface between nodule and concrete, e is the thickness of concrete
Table 2
Evaluation of the contact pressure vs lime and concrete properties.

Fig. 6. Connement of the free lime nodule inside the concrete.

Cellular concrete block C2


Cellular concrete block C4
Concrete

EC
(MPa)

R/Rmean
(%)

R/Rmax
(%)

Pmean
(MPa)

Pmax
(MPa)

2000
4000
15,000

22
16
7

24
19
12

32
24
11

35
29
19

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

covering and is the angle of the ejection cone with respect to the concrete surface. Eq. (9) simply translates the fact that the resultant of
swelling pressure acting on the bottom part of the nodule is directly balanced by the vertical resultant of tensile stresses acting at the surface of
the ejection cone and that the pop-out effect is initiated when the local
tension stresses exceed the tension resistance of the concrete matrix.
Based on the evaluation of the swelling pressure obtained from
model 1, tensile stress can be calculated from model 2 and compared
to the tensile strength of concrete to assess the risk of pop-out.

Table 3
Evaluation of the contact pressure vs E modulus of lime (EL) and concrete (EC) and volume
(radius) variation of lime.
EL
(MPa)

V/V
(%)

R/R
(%)

p [EC = 7000]
(MPa)

p [EC = 15,000]
(MPa)

p [EC = 30,000]
(MPa)

100

10
20
30
40
50
10
20
30
40
50
10
20
30
40
50

3
6
9
12
14
3
6
9
12
14
3
6
9
12
14

6
12
18
23
28
9
18
27
35
42
12
24
35
46
56

6
12
18
24
29
10
19
27
35
43
13
25
36
47
57

6
12
18
24
29
10
19
27
35
43
13
25
36
47
57

150

3. Description of the experimental programme


The objectives of the experimental investigations and analyses
were:

to observe the behaviour of a lime nodule embedded into concrete,


to quantify expansion rate of lime,
to characterize the quality of hydrated lime (densication process),
to select data for calculation modelling (expansion and compressibility
modulus of free lime).

3.1. Selection and preparation of materials


Free lime (from 80 to 120 mm limestone blocks) is representative
of a production process with a classical reactivity (T60 = 2 min 17 s)
and purity; density is equal to 1.44 g/cm3. Samples ( = 18.5 mm
and H = 15 to 20 mm) are cored in the lump lime. Concrete blocks
of 190 190 85 mm3 have been rstly used to simulate conning
effect. They provide a compressive strength of 15 N/mm2 and a density
of 2.2 g/cm3. This type of porous concrete has been selected in order to
favour water transmission and lime reaction.
3.2. Conning operations
A cylindrical opening of 19 mm diameter (Fig. 5a) and 25 mm deep
(Fig. 5b) is cored in the blocks by means of a drill. To study the impact of
the situation of the nodule of lime from the concrete surface, the holes
are made at different distances from the edge of the concrete block
(Fig. 5c). To ensure the best connement possible, nishing cylindrical
orices are performed with quick-setting cement to ll the cone left
by the bit, to smooth the cylinder walls and to repair the damaged
edge of the whole drilling. However, despite these precautions, connement is not perfect: the top surface of the block is not perfectly at and a

77

200

minimum clearance is required to introduce the carrot in his hole


(Fig. 5d).
The connement of the concrete block is then ensured by steel plates
and clamps. A plastic sheet is inserted under the steel plate for
preventing any reaction between lime and steel (Fig. 6). The assembly
is then immersed in water and the water level is adjusted under the
top edge of the block (Fig. 7).
A reference test is made in order to check the time needed for a
complete hydration of a free lime cylinder: it was measured that
67 hour storage into water allowed a full hydration of the sample by
means of water transfer through the porosity of the sample.
The blocks are then cut with a diamond saw (dry) at the right and
the edge of the nodule, in order to observe possible internal cracks. Hydrated lime cores are then recovered, measured, weighed and analyzed
on the base of the loss on ignition at 100 C (moisture measurement)
and 600 C (measuring the rate of hydration).
3.3. X Ray Diffraction and microscopical analysis
X Ray Diffraction analyses have been carried out in order to determine
the mineralogy of the crystals and the nature of the hydrated products.
Samples observed with Environmental Scanning Electron Microscope
(ESEM) are similar to the one used for XRD investigations. The samples
are glued on metal pads using a conductive adhesive. The specimens are

Fig. 8. Comparison of the actual and assumed pressure/volume relationship.

78

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

Fig. 9. Evolution of compressive stresses vs contact pressure and nodule diameter.

thus metallized with Pt before being introduced into the vacuum chamber
of a scanning electron microscope ESEM. The electron beam gives a view
of topography and shape of hydrated lime. EDAX (Philips) system,
coupled with ESEM, allows the detection of elements identied on a
spectrum according to their energy dispersion.
These analyses allow obtaining a good identication of different
forms of calcium hydroxide crystals under connement.
4. Results and discussions
4.1. Stress calculation and modelling hypothesis
In order to feed the proposed models, data needed are:
Elastic modulus of hydrated lime and concrete,
Poisson's ratio of hydrated lime and concrete,
Variation of volume of nodule during hydration.
The elastic modulus and Poisson's ratio of lime are estimated from
[20]: the value used for the elastic modulus of 100 MPa and Poisson's
ratio is 0.25. Results obtained from the experimental programme running

in parallel with the present theoretical study indicate similar values


(chapter 4). The module of the concrete is of the order of 30 GPa and its
Poisson's ratio is considered equal to 0.2. It will be shown later on that
the results are quite insensitive to these latter values.
The swelling rate is more difcult to estimate. In fact, if the volume
variation of the lime during hydration can reach values as high as
200%, these values are only relevant in the absence of any connement.
The tests carried out in the framework of the present study show indeed
that the volume variation signicantly depends on the stiffness and
strength of the conning medium. The results presented in chapter 4
and other tests on lightweight concrete blocks show that, if one measures the change in diameter of cylindrical samples in the direction of
connement (i.e. in the radial direction) for samples with a diameter
which is adjusted to the initial hole made in the conning medium,
and which therefore are in contact with this medium from the start of
the hydration reaction, the average value of R/R is:
22% for connement in cellular concrete block of 2 MPa compression
characteristic strength,
16% for connement in cellular concrete block of 4 MPa compression
characteristic strength,

Fig. 10. Evolution of tensile stresses vs contact pressure and nodule diameter.

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388


Table 4
Thickness of the potentially cracked zone around nodule of D diameter vs tensile strength.
V/V
(%)

fct = fctC25/30,5% = 1.8


(MPa)

fct = fctC25/30,50% = 2.6


(MPa)

fct = fctC25/30,95% = 3.3


(MPa)

10
20
30
40
50

0.19 D
0.36 D
0.48 D
0.57 D
0.64 D

0.11 D
0.26 D
0.37 D
0.45 D
0.51 D

0.07 D
0.21 D
0.30 D
0.37 D
0.43 D

7% for connement in a heavy concrete block of 15 MPa compression


characteristic strength.
These values correspond to respective volume variations (V/V) of
49, 34 and 15%, respectively.
This observation is explained by the fact that, when the conning
pressure reaches a sufcient level, it forces the hydration reaction to
occur towards the inside of the lime sample. This reaction proceeds
thus at constant volume and pressure, but results in a more dense
hydrated material. It is therefore expected that its elastic modulus be
higher than the reference value of 100 MPa.
Eqs. (7), (8) and (9) cannot be directly applied in the purpose of
comparison with the case of cylindrical test samples: the above
equations indeed assume a spherical inclusion. The transition to the
cylindrical case, however, can be done quite simply by changing the
coefcients Cb and Cch:
12L 1 L
CL
EL

10

and

CC

1 C
:
EC

11

Taking into account this adaptation, it is possible to evaluate contact


pressure developed for the different test conditions (Table 2).

79

One might conclude from the values in Table 2 that the pressure required to conne the reaction varies with the material. It must however
be noted that the model assumes an elastic behaviour of the conning
material whatever the value of the pressure, while an evaluation of
the stresses developed in the conning medium for the cellular concrete
block case (i.e. c,max = p and t,max = p/2) concludes that they exceed
by far the resistance of the blocks. Calculations show that, for a sample
of 18 mm diameter, the resistance is actually exceeded on a thickness
approximately equal to the diameter of the sample. It can therefore be
estimated that the effective stiffness of the zone of the conning
medium directly surrounding the inclusion is reduced with respect to
its reference undamaged value, thereby increasing the value of the
coefcient CC and reducing consequently the pressure so as to reach a
balanced situation between effective stiffness and pressure. A rened
modelling of this phenomenon would however require advanced tools
that are considered out of the scope of the present study. One must
also note that cracks were indeed observed in the zone surrounding
the inclusion for some of the cellular concrete blocks used for connement tests, which is in accordance with the above conclusions. On the
other hand, for an ordinary concrete containment, the calculated level
of pressure is such that the stress remains at a level below the resistance
of the material. The model reproduces thus correctly the experimental
trends.
The range of parameters considered for the upcoming parameter
studies is dened by:
EC: 7000, 15,000 and 30,000 MPa, which corresponds to concrete
connement ranging from poor characteristics to ordinary
concrete;
EL: 100 to 200 MPa, ranging from a normal value to a value doubled
to take into account densication of lime during hydration. Experimental tests shows actually that, for the samples corresponding to
hydration in heavy concrete blocks, and thus with the highest
densication rate, the value of the measured elastic modulus is
about 150 MPa;
L and C are equal to 0.25 and 0.20 (values from the technical
literature), respectively;
Change in volume of the spherical inclusion varying from 10 to
50%, which corresponds to changes in radius of 3 to 15%, respectively. This range roughly sweeps the radius variation observed

Fig. 11. Evolution of the minimal thickness of concrete cover vs nodule diameter (minimum pressure).

80

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

Fig. 12. Evolution of the minimal thickness of concrete cover vs nodule diameter (mean pressure).

during tests on hydration inside the heavy concrete blocks


(observed values varying between 3 and 12%).
Finally, it must be stated that, as illustrated in Fig. 8, the pressure
level provided by the model for a given volume variation (Pn,model) is
obviously an upper bound of the actual value (Pn,actual) due to the
following reasons:
The model assumes that the inclusion and the conning materials are
in perfect contact and exhibit a spherical symmetry, while they
actually very often show signicant shape irregularities. In order to
provide order of magnitudes, it is relevant to note that, for same
volume variation and initial diameter, the pressure calculated for a
cylindrical inclusion is 25% less than the one calculated for a spherical
inclusion (i.e. the slope of the real pressure vs volume curve is
lower than the one given by Eqs. (3) and (6));
The model assumes that the contact is established from the beginning
of swelling. In reality, it can be assumed that a part of the measured
volume variation (V0) is corresponding to the lling of the voids

between the inclusion and the matrix; the real pressure should
therefore be associated with only a fraction of the volume change;
The model assumes that the elastic modulus of lime is constant
throughout the swelling phase, whereas it is in fact a phenomenon
whose parameters vary with time. The mechanical characteristics of
hydrated lime at the initiation of the swelling (ini) are corresponding
to an unconned environment and progressively evolve to those of
hydrated lime properties in conned environment (fin). The model
conservatively considers the stiffer situation (fin) throughout the
entire swelling process.
The level of conservatism of the chosen modelling assumptions is
however quite impossible to quantify.
4.2. Model exploitation and parametric study
4.2.1. Evaluation of the contact pressure
Results of the calculation of the contact pressure are presented on
Table 3. It is showed that pressure is roughly independent from the

Fig. 13. Evolution of the minimal thickness of concrete cover vs nodule diameter (maximum pressure).

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388


Table 5
Hydrated lime content of the samples.
Test

Loss of ignition at 600 C


[%]

Ca(OH)2 content
[%]

8
9
12
17

24.2
23.9
23.6
23.7

99.6
98.1
96.9
97.4

quality of concrete. However, results are clearly inuenced by the rigidity of the lime and the variation of the volume: estimated pressure
ranges from 6 to 57 MPa (this latter value is clearly unrealistic but
denes an absolute upper bound that would never be overtaken even
for the worst conditions). It has also to be mentioned that a possible variation of the Poisson's ratio of the lime could induce signicant variation

(a) Free lime nodule at 10 mm depth

(c) Free lime nodule at 40 mm depth

81

of the pressure: e.g. for an elastic modulus equal to 100 MPa, a variation
of volume corresponding to 30% and Poisson's ratio equal to 0.2, 0.25
and 0.3, resulting pressure is equal to 15, 18 and 23 MPa respectively.
For the reasonable average values of the parameters (EL = 150 MPa
and volume variation of 20%), pressure is up to 19 MPa, which is however
a clearly conservative estimate of the real pressure, as illustrated on Fig. 8.
4.2.2. Stress distribution
Compressive and tensile stress distributions are given in Figs. 9 and
10 for several values of pressures and nodule diameters. Curves are
corresponding to 3 levels of pressure taken from Table 3:
Pmin = 6 MPa (pressure calculated on the base of minimalistic
hypotheses),
Pmean = 19 MPa (pressure conservatively calculated on the base of
most probable hypotheses),

(b) Free lime nodule at 20 mm depth

(d) Free lime nodule at 90 mm depth

(e) Free lime nodule at 5 mm depth : pop out but no internal crack
Fig. 14. Sections of concrete blocks and free lime nodules.

82

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

(a)

(b)

Fig. 15. Free lime nodule at 5 mm depth (a) and 10 mm depth (b).

Pextr = 57 MPa (pressure upper bound calculated on the base of


extreme hypotheses).
Values of stresses are calculated for 3 lime nodule diameters: 2, 10
and 20 mm, respectively. On Fig. 9 providing tensile stresses, horizontal
straight line corresponds to the mean tensile strength of ordinary
C25/30 concrete.
The gure related to compressive stresses (Fig. 9) shows that,
around a nodule which is assumed to be perfectly smooth and spherical, the stress level is equal to the pressure. For extreme conditions,
this level is likely to exceed the level of resistance of ordinary concrete. However, for average conditions, this level remains acceptable.
The curves also clearly show that the stress level rapidly decreases
with the distance from the nodule. Consequently, at a distance
equal to two times the diameter, the stress level is only about 1% of
the pressure and goes down to less than 0.1% at a distance
corresponding to 5 times the diameter. It can therefore be concluded
that:
Local crushing of the concrete near the nodule is normally not to be
feared;
The area of inuence of a nodule is of the order of 2 to 3 times the diameter (measured from the centre of the nodule). Assuming that the
minimum distances between nodules are in the order of 100 mm, the
areas potentially impacted by the presence of nodules are unlikely to
interfere.

Fig. 10 shows that, except for the minimum conditions, there exists a
region around the nodule where tensile forces are potentially present
and thus likely to initiate micro-cracking. No signicant cracking has
however been observed in tests performed in the laboratory for lime hydration in heavy concrete blocks (chapter 4). Table 4 gives the thickness
of the potentially cracked area corresponding to a module of lime E =
150 MPa for different values of the volume change and of the tensile
strength of the concrete (corresponding to percentiles of 5, 50 and 95,
respectively, for standard concrete C25/30).
One must be reminded that the values of radial swelling measured
in the laboratory (chapter 4) are of the order of 7% (so corresponding
to a volume variation V/V about 15%). However, in the worst case
(V/V = 50%) and for a very low concrete tensile strength (1.8 MPa),
the thickness of the cracked area is potentially of the order of 64% of
the diameter. Under these conditions, if these cracked zones are fully assimilated to non-resistant inclusions with a diameter equivalent to 2.28
times the diameter of the nodule (i.e. 1 D + 2 0.64 D), the calculation
of the effect of voids on the compressive strength of concrete according
to Section 1 of the present paper shows that, even in the case of cracking
around all the nodules included in the concrete (for a reasonable concentration of such lime nodules), the overall strength of the concrete
would hardly be affected.
4.2.3. Pop-out risk estimation
Figs. 11, 12 and 13 represent the results obtained by using the Eq. (9)
to estimate the risk of pop-out. These gures actually represent the

Table 6
Measurements of conned lime cylinders in concrete blocks (D is the distance between the nodule and the surface and d is the diameter of the nodule).
Test

Maximal connement

Restricted connement

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

Initial lime

Observations

Characteristics of hydrate

Density
(g/cm3)

D/d

Pop-out
(Y/N)

Cracking
(Y/N)

Density
(g/cm3)

Volume expansion
(%)

Radial expansion
(%)

Longitudinal expansion
(%)

1.38
1.58
1.44
1.44
1.38
1.49
1.47
1.39
1.43
1.40
1.41
1.46
1.50
1.39
1.45
1.51
1.41
1.50

4.5
4.5
4.5
4.5
4.5
4.5
2
2
1
1
1
0.5
0.5
0.5
0.25
0.25
2
0.25

N
N
N
N
N
N
N
N
N
N
N
Y
N
N
Y
Y
N
N

N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N

1.39
1.71
1.62
1.67
1.47
1.54
1.66
1.50
1.49
1.55
1.52
Unrecovered
1.50
1.46
Unrecovered
Unrecovered
1.14
1.31

32
22
18
14
24
28
17
23
27
20
23

10
5
7
3
6
8
8
6
8
8
6

8
11
4
6
11
9
1
8
8
3
9

32
25

12
7

5
10

64
51

24
27

76
73

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

83

1.84
Fig. 16. Evolution of the density of hydrated lime vs volume expansion (conned (tests 1 to 16) and less conned (tests 17 and 18)).

minimum thickness of the concrete cover beyond which the theoretical


model predicts no occurrence of pop-out. This thickness is plotted as a
function of the diameter of the inclusion for different values of the
tensile strength of concrete and of the swelling pressure of the nodule.
The ejection angle is chosen equal to 30 in accordance with common
observations made on site. For facilitating the reading of the charts, a
light grey dash line outlines the situation where concrete thickness
equals the diameter of the nodule.
The following observations can be drawn from the analysis of these
gures:
For the minimal assumptions (low elastic modulus of the lime and
low ination rate Fig. 11), the risk of pop-out is almost negligible.
The model predicts occurrence of pop-out only for very low quality
of concrete (fct equal to 1.8 MPa). For higher values of fct, the estimated risk of pop-out is zero whatever the cover thickness, as shown on
Fig. 11 by perfectly horizontal superimposed lines corresponding to
fct equal to 2.6, 3.3 and 5.0 MPa. Even in the worst case, a cover
thickness less than 10% of the diameter appears sufcient to prevent
pop-out;
For the most realistic though conservative assumptions (intermediate
modulus of the lime and swelling ratio of 20% in volume Fig. 12), the
risk of pop-out is more important. For a tensile strength corresponding to ordinary concrete (2.6 MPa), the risk of pop-out occurs if the
thickness of the covering is of the order of half of the diameter (50%)
of the inclusion. This is to be compared with hydration tests on

Table 7
Rigidity modulus and compressive strength of hydrated lime cylinders (MPa).
Specimen reference
a

13 centred
14 centredb
17 centred
2C20
4C10
5C40
10C10
a
b

Small crack at mid-height.


Non parallel faces.

Compressive strength
[MPa]

Rigidity modulus
[MPa]

5.1
5.1
5.4
8.3
3.8
5.3
5.4

150
164
189
168
129
180
144

heavy concrete blocks (inclusions approximately 20 mm diameter)


for which pop-outs have been observed only for cover thicknesses of
10 and 5 mm (i.e. 50 and 25% of the diameter). Modelling and
observations seem to provide convergent results, conrming thus
the assumptions made on the different parameters entering in the
modelling process;
For extreme cases (Fig. 13), the model predicts ejection of pop-outs
for thicknesses of less than 1 to 2 times the diameter, according to
tensile strength of the concrete.
4.3. Conned free lime hydration results
4.3.1. Results of the tests
Eighteen trials were conducted, including 16 in the form of maximum connement (tests 1 to 16) and two under restricted connement
(tests 17 and 18), leaving more place for the hydrate to crystallize: the
connement is in fact restricted if the carrot of lime is of a size smaller
than the hole in the concrete. In this case, the metal plate is not
completely in contact with the concrete surface and allows some expansion of the lime. Several distances D from the edge of the concrete
substrate were tested: 4.5 to 0.25 times the initial diameter d of the
core of lime, respectively D/d in Table 6.
After testing conditions 8, 9, 13 and 17, respectively, loss of ignition
at 600 C was measured in order to evaluate the rate of hydration of the
lime (Table 5).
A minimum value of 97% of Ca(OH)2 is measured. If we consider that
initial lime is not totally pure it means less than 100% CaO due to
unburn limestone we may conclude that all the free lime has been
hydrated.
4.3.2. Internal cracking of the concrete blocks and pop-outs
No cracking has been observed for all the concrete blocks tested
(Fig. 14 a to e). Even for free lime nodules very close to the concrete surface (0.5 times free lime nodule diameter), no cracking appeared inside
the block.
Pop-outs were observed only for specimens (Fig. 15a and b) corresponding to depths:
0.25 times free lime nodule diameter (2 by 3 of the specimens) and
0.5 times free lime nodule diameter (1 by 3 of the specimens).

84

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

Lin (Counts)

Portlandite
220
210
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
0

NC1

d=2.62351

Portlandite
d=4.89551

Portlandite
d=3.10515

d=23.80000
d=26.83630d=16.23526

10

20

30

40

2-Theta - Scale
NC1 - File: NC1.raw - Type: 2Th/Th locked - Start: 2.000 - End: 45.000 - Step: 0.020 - Step time: 0.6 s - Temp.: 25 C (Room) - Time Started: 8 s - 2-Theta: 2.000 - Theta: 1.000 - Chi: 0.00 - Phi: 0.00 - X: 0.0
Operations: Background 0.000,0.000 | Smooth 0.150 | Import
04-0733 (I) - Portlandite, syn - Ca(OH)2 - Y: 86.46 % - d x by: 1. - WL: 1.5406 - 0 - I/Ic PDF 1.4 -

Fig. 17. XRD analysis on NC1 sample.

When depth was higher than 1 times the free lime nodule diameter,
no pop-out was observed.
4.3.3. Analysis of densication phenomenon and nodule expansion
As described in 3.2, the blocks are cut with a diamond saw (dry) at
the right and the edge of the nodule and hydrated lime cores are then
recovered, measured and weighed. As a reminder, in unconned
conditions, when quicklime CaO (bulk density about 1.5 g/cm3) is
transformed into Ca(OH)2 hydrate, it is in the shape of a powder of
low density (about 0.5 g/cm3). The measurements show (Table 6), in
conned situation, a densication process of the hydrate which
offers a density much higher than in unconned case. During these
tests, the hydrates obtained in conned environment showed a density
varying from 1.4 to 1.7 g/cm3, with an average value of 1.55 g/cm3
(Table 6).
This densication process logically implies a volume expansion
factor much lower than what is generally known when unconned

(300% as calculated from the ratio of 1.5 g/cm3 for lime divided by
0.5 g/cm3 for Ca(OH)2 hydrated lime). Expansion factors measured
here are in the range from 15 to 30% (23% average). Fig. 16 shows the
experimental values in the case of a high conned situation (tests 1 to
16, rhomb dots) and in the case of a partial connement (tests 17 and
18, square dots).
The positive expansion that is observed in the case of conned
samples (rhumb dots) means that these connements are not perfect,
leaving a free space for expansion. This imprecision is statistically distributed across different samples. These expansions are not related to
a deformation of the concrete block. It is possible to estimate the density
of the core hydrate in the case of perfect connement by extrapolating
the regression line. This extrapolation (82% correlation) allows to
calculate that, in the case of perfect connement (no volume expansion), the hydrate, formed from lime industrial bulk with a density of
1.5 g/cm3, will have an equivalent density of 1.84 g/cm3: this remains
less than the absolute density of the hydrate (2.2 g/cm3).

Fig. 18. ESEM observations on NC1 sample.

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

85

Fig. 19. ESEM observations on C110 sample.

The tests allow concluding that hydration of quicklime in a conned


environment, leads to the production of completely hydrated portlandite
nodules with very high densities.

4.3.4. Rigidity modulus


Some values from 100 to 200 MPa are given in the literature [19].
Due to the intrinsic variability of the free lime, tests have been performed on cylinder used for testing connement effect. Samples are
prepared exactly in the same conditions than in 3.2: after 67 hour
hydration, the cylinders are cored from concrete blocks (20 mm and
H1520 mm). Until testing, specimens are stored into plastic bags in
order to avoid carbonation process. Compressive loading is applied at
a speed of 5 N/s on an Instron 5585 tensile machine (Table 7).
These indicative measurements conrm literature results and have
been used as reference values for modelling (chapter 3).

On diffractograms (Fig. 17), it appears that all the samples are of


portlandite type Ca(OH)2. Calcite can be present in very few quantities,
due to carbonation.
4.3.6. ESEM observations
The EDAX analysis conrmed the XRD analysis with bands, next to
oxygen and platinum bands, characterized by the presence of Ca.
4.3.7. NC1 sample
The unconned sample is in the form of powder, hard to stick on the
pad and difcult to be metallized. The porosity (Fig. 18(a)) is very high.
The grains are generally anhedral, rarely euhedral. They have a size
of 1 to 5 m (Fig. 18(b)).

4.3.5. Hydrated lime analysis


X Ray Diffraction analyses have been carried out in order to determine the mineralogy of the crystals and the nature of the hydrated
products. Several samples have been recorded:

4.3.8. C110 sample


The sample shows expansion cracks (Fig. 19(a)). As in the periphery
than in the centre of the sample, small euhedral crystals are observed;
they attest that there was free space for them for growing and adopting
their own crystalline form. The crystals are associated with very small
needles blooms (Fig. 19(b)). The size of portlandite crystals ranges
from 0.1 to 5 m.

NC1: not conned sample (100% free volume Vf)


C110: 10 mm sample conned into concrete hole of 20 mm
diameter (75% Vf)
C115: 15 mm sample conned into concrete hole of 20 mm
diameter (44% Vf)
C120: 20 mm sample conned into concrete hole of 20 mm
diameter (0% Vf)

4.3.9. C115 sample


Like for sample C110, the edge of the core is characterized by the
presence of expansion cracks. As in the periphery than in the centre,
small euhedral crystals are visible (Fig. 20(a)), which attest about the
free space that has existed around them for developing and adopting
their crystalline shape. The size of portlandite crystals ranges from 2
to 5 m (Fig. 20(b)).

Fig. 20. ESEM observations on C115 sample.

86

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

Fig. 21. ESEM observations on C120 sample.

4.3.10. C120 sample


The puck taken from the sample is very dense. It is cut in the core
using an instrument that has left its mark: the wall of the core is regular
and detached without tearing (Fig. 21 sample C120). Anhedral grains
are joined and the porosity is low (Fig. 21(a)). The grain size is about
1 m (Fig. 21(b)).
4.4. Comparisons and analysis of observations
At a low magnication (Fig. 22), two samples show common features (the expansion cracks): C110 (Fig. 19(a) and C115 (Fig. 22(b)).
The unconned NC1 sample (Fig. 22(a)) is in powder form while C120
appears to be more massive (Fig. 22(c)).
At a higher magnication (1000 to 3000 ), even if NC1
(Fig. 18(b)), C110 (Fig. 19(b)), C115 (Fig. 20(b)) samples are similar

(a) NC1

in grain size, this is not the case with regard to their crystallinity: only
C110 and C115 are characterized by the presence of euhedral crystals
(Fig. 23). Grain size is ner for sample type C120.
The scanning electron microscopy allowed observing the structures
of lime processed under various conditions of connement, both on site
and in the laboratory. The comparison of samples generated in the laboratory shows the difference between unconned and conned, and, for
conned samples, the evolution of structures versus rates of expansion.
Structure and porosity may be compared with the densities of cylinders
(Table 6).
5. Conclusions
A simple model based on the theory of elasticity, on duly validated
experimental data and on reasonable engineering judgement has been

(b) C115

(c) C120
Fig. 22. Comparison of the samples under ESEM (low magnication).

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

(a) NC1

(b) C110

(c) C115

(d) C120

87

Fig. 23. Comparison of the samples under ESEM (high magnication).

derived to estimate the consequences of the presence of hydrated lime


nodules from quicklime on the mechanical characteristics of structural
concrete and on the risk of occurrence of pop-outs likely to inuence
concrete durability. By exploiting this simple model, the following
conclusions can be drawn:
for the range of parameters that have been selected, the pressure
developed at the interface between the concrete and the hydrated
nodule varies from 6 to 60 MPa. The most probable value is of the
order of 20 MPa;
the area of inuence of a nodule is of the order of 2 to 3 times its
diameter;
no local crushing of concrete in the vicinity of a nodule is to be considered. If such a crushing would anyway occur, it would only affect a
very limited zone around the nodule with no impact on the overall
mechanical properties of the concrete;
a micro-cracked zone is likely to develop around the nodule under the
conjunction of unfavourable conditions. The diameter of this cracked
region could be at most of the order of 230% of the diameter of the

nodule and its impact on the strength of the concrete at the macroscopic level is proven as negligible;
for the most probable values of the swelling pressure, the minimum
concrete cover thickness allowing the prevention of the pop-out
phenomenon is of the order of half the diameter of the inclusion. In
other words, under the assumptions considered in this study, no
nodule located at a depth of more than half its diameter should
cause pop-out even when hydrated;
The following conclusions can be drawn from the experimental
programme:
hydration of quicklime in a conned environment, leads to the
production of completely hydrated portlandite nodules;
hydrate formed in a conned environment occupies the available volume and, in present cases, may reach very high densities (average
value = 1.55 g/cm3);
the rate of volume expansion of the quicklime, depending on the free
volume, is very low (average 23% expansion) compared to an
unconned and is the result of a non-full connement;

88

L. Courard et al. / Cement and Concrete Research 64 (2014) 7388

in the case of a fully enclosed environment (case of lime nodules


trapped in a concrete structure), the density of the hydrate may
reach up to 1.84 g/cm3, which is still under 2.24 g/cm3 (absolute
and therefore maximum density);
pop-out appears in the test conditions for depths less than or equal to
0.5 times the initial diameter of the nodule of lime;
no internal cracking is observed in the concrete blocks;
when connement is maximal, anhedral grains are joined and the
porosity is low; however, when space around nodules was available,
porosity is large and grain shape is euhedral.
Finally, laboratory tests clearly show that the depth of connement
is the most important factor for explaining pop-out and free lime expansion. Moreover, just the near-to-surface layer is affected by the risk of
pop-out: when the nodule is under the concrete surface, the
surrounding concrete is sufciently resistant to conne the nodule
and avoid explosion.
Under the worst case scenarios in combined terms of swelling
pressure and concrete strength, the minimum thickness necessary to
prevent the pop-out phenomenon is of the order of two times the
diameter of the inclusion. In other words, even under these extremely
unfavourable assumptions, no nodule located at a depth of more than
2 times its diameter should cause pop-outs.
Synopsis
Pop-out induced by lime expansion is a well-known phenomenon.
The concrete cover thickness, the diameter and the shape of the lime
nodule as well as the mechanical characteristics of concrete and lime
are the key parameters inuencing the development of internal pressure and hence controlling the risk of cracking or pop-out. Laboratory
investigations and modelling have been performed and show that the
minimum concrete cover thickness necessary to avoid the development
of the pop-out phenomenon is estimated in the order of half the diameter of the inclusion.

References
[1] C. Plinius Secundus (maior), Natural history, translation from Latin by E. Littr (2nd
Tome), Paris, 1850.
[2] Ph. Dumont, Use of lime in construction, CERES 99/1, University of Lige, Belgium,
1999. (in French).
[3] J.A.H. Oates, Lime and limestone, Chemistry and technology. Production and uses,
Wiley-VCH verlag GmbH, Weiheim, 1998.
[4] R.S. Boynton, Chemistry and Technology of Lime and Limestone, 2nd edition, Wiley
& Sons, New York, (578 pages).
[5] S. Chatterji, Mechanism of expansion of concrete due to the presence of dead burnt
CaO and MgO, Cem. Concr. Res. 25 (1) (1995) 5156.
[6] M.H. Lee, J.C. Lee, Study on the cause of pop-out defects on the concrete wall and
repair method, Constr. Build. Mater. 23 (1) (2009) 482490.
[7] G.M. Idorn, Expansive mechanisms in concrete, Cem. Concr. Res. 22 (1992)
10391046.
[8] A. Verhasselt, Industrial by-products for the design of bonded layers in foundations:
blast furnace slags, iron slags and y ashes, Belgian Road Research Center
Publication, CR 33/91, 1991. (in French).
[9] F. Choquet, Laboratory study for the valorization of iron slags and blast furnace slags
in road engineering, Belgian Road Research Center Publication, CR 22/84, 1984.
(in French).
[10] M. Deng, D. Hong, X. Lan, M. Tang, Mechanism of expansion in hardened cement
pastes with hard-burnt free lime, Cem. Concr. Res. 25 (2) (1995) 440448.
[11] L. Mun-Hwan, L. Jong-Chan, Study on the cause of pop-out defects on concrete wall
and repair method, Constr. Build. Mater. 23 (1) (2009) 482490.
[12] B. Chiaia, A. Fantilli, G. Ferro, G. Ventura, Modeling the CaO hydration in expansive
concrete, in: Bicanic (Ed.), Computational Modelling of Concrete Structures, CRC
Press, 2010, pp. 441449.
[13] H. Shi, L. Yuan, Theoretical and experimental research on expansive stress in
hardened cement paste, Adv. Concr. Res. 4 (2004) 155160.
[14] C.F. Dunant, K.L. Scrivener, Effects of uniaxial stress on alkalisilica reaction induced
expansion of concrete, Cem. Concr. Res. 42 (2012) 567576.
[15] J. Faury, Concrete: inuence of inert components requirements for a better design,
Dunod Ed., Paris, 1942. (in French).
[16] G. Dreux, New guide for concrete, 7th edition Eyrolles Ed., Paris, 1995. (in French).
[17] R.D. Hooton, A.M. Ramezanianpur, R.M. Ahani, U. Schutz, Durability of Portland
limestone cement concrete, International Congress on Durability of Concrete.
ICDC, Trondheim, Norway, 2012. (10 pp.).
[18] B. Chiaia, A. Fantilli, G. Ventura, A chemo-mechanical model of lime hydration in
concrete structures, Constr. Build. Mater. 23 (2012) 308315.
[19] S. Timoshenko, J.N. Goodier, Theory of elasticity, 2nd edition McGraw-Hill Book
Company, 1951.
[20] K. Van Baelen, Carbonation of lime mortars and its effect on historical structures,
(PhD thesis) Catholic University of Leuven (KULeuven), 1991. (in Dutch).